Lie Algebras and Lie Groups Homework VIII

Pádraig Sheehy
Due: 15-11-2019

Problem 1
Let V (n) denote the unique n + 1 dimensional irreducible representation of sl2 (k). Show that
1. The trivial representation k is isomorphic to V (0).
2. The standard representation k2 is isomorphic to V (1).
3. The adjoint representation sl2 (k) is isomorphic to V (2).

Find the decomposition of these representations into a direct sum of h-eigenspaces.

Solution:
V (n) denotes the unique n + 1 dimensional irreducible representation of sl2 (k). Thus, it is only necessary to
show that the given representations are irreducible.
1. The trivial representation k is a one-dimensional representation and is thus irreducible.
So, V (0) ' k.
2. The standard representation k2 is irreducible, by Homework 7 Problem 4.
So, V (1) ' k2 .
3. By Example 3.5 and 3.39, we know that sl2 (k) is simple. So, by Remark 3.58, we have that the adjoint
representation V = sl2 (k) is irreducible. So, V (2) ' sl2 (k).
Having shown this, we can directly apply Theorem 4.1:

V (n) = V−n ⊕ V−n+2 ⊕ . . . ⊕ Vn−2 ⊕ Vn ,

where Vk is the subspace of h-eigenvectors with eigenvalue k. So we get:

1. V (0) ' k =⇒ k = V0
2. V (1) ' k2 =⇒ k2 = V−1 ⊕ V1 .
3. V (2) ' sl2 (k) =⇒ sl2 (k) = V−2 ⊕ V0 ⊕ V2 .

Problem 2
Find the decomposition of sl2 -representations
1. V (2) ⊗ V (2),

2. Hom(V (1), V (2))

into a direct sum of irreducible representations.

1
Solution:
1. Recall if V, W are L-modules, then V ⊗ W is an L-module where x(v ⊗ w) = xv ⊗ w − v ⊗ xw.
Let v ∈ Vi , w ∈ Vj , we get:

h(v ⊗ w) = hv ⊗ w − v ⊗ hw
= i(v ⊗ w) + j(v ⊗ w)
= (i + j)(v ⊗ w)

By Theorem 4.1, we have V (2) = V−2 ⊕ V0 ⊕ V2 , and so we get:

V−2 ⊗ V−2 → −4 V0 ⊗ V−2 → −2 V2 ⊗ V−2 → 0

V−2 ⊗ V0 → −2 V0 ⊗ V0 → 0 V2 ⊗ V 0 → 2
V−2 ⊗ V2 → 0 V0 ⊗ V2 → 2 V2 ⊗ V2 → 4

Let mk denote the multiplicity of V (k) in the decomposition.

Applying Remark 4.4, we get:

m4 = 1 − 0 = 1
m3 = 0 − 0 = 0
m2 = 2 − 1 = 1
m1 = 0 − 0 = 0
m0 = 3 − 2 = 1

Thus V (2) ⊗ V (2) = V (4) ⊕ V (2) ⊕ V (0)

2. Recall if V, W are L-modules, then Hom(V, W ) is an L-module where xf (v) = x · f (v) − f (xv).
Let v ∈ Vi , f ∈ Hom(Vi , Vj ). We get:

(hf )(vi ) = h(f (vi )) − f (h(vi ))

= jf (vi ) − f (ivi )
= jf (vi ) − if (vi )
= (j − i)f (vi )

Applying Theorem 4.1, we get V (1) = V−1 ⊕ V1 and V (2) = V−2 ⊕ V0 ⊕ V2

So examining the weights of the combinations, we get:

Hom(V−1 , V−2 ) → −1 Hom(V−1 , V0 ) → 1 Hom(V−1 , V2 ) → 3

Hom(V1 , V−2 ) → −3 Hom(V1 , V0 ) → −1 Hom(V1 , V2 ) → 1

Applying Remark 4.4, we get:

m3 = 1 − 0 = 1
m2 = 0 − 0 = 0
m1 = 2 − 1 = 1
m0 = 0 − 0 = 0

Thus Hom(V (1), V (2)) = V (3) ⊕ V (1)

Problem 3
The algebra of polynomials A = k[u, v] has the Lie algebra of derivations
∂ ∂
DerA = {f ∂u + g∂v | f, g ∈ k[u, v]}, ∂u = , ∂v = .
∂u ∂v
Show that:

2
 1. The linear map
ρ : sl2 → DerA ⊂ gl(A) : x 7→ u∂v , y 7→ v∂u , h 7→ u∂u − v∂v
is a Lie algebra homomorphism. Hence (A, ρ) is a representation of sl2 .
2. The subspace k[u, v]n ⊂ k[u, v] of homogeneous degree n polynomials is a subrepresentation of dimension
n + 1.
3. The linear map
1
φ : V (n) → k[u, v]n : vi 7→ un−i v i
(n − i)!
is an isomorphism of sl2 -representations.
The basis (vi )ni = 0 of V (n) was defined in the lecture; it satisfies hvi = (n − 2i)vi , xv0 = 0, vi = y i v0 .

Solution:
1. We’re given that the map is linear, so it remains to show that ρ respects the bracket. It suffices to check
this for each pair of distinct basis elements.
[ρ(x), ρ(y)] = u∂v (v∂u ) − v∂u (u∂v )
= u∂u + uv∂u ∂v − v∂v − uv∂u ∂v
= u∂u − v∂v
= ρ(h)
= ρ([x, y]),

[ρ(x), ρ(h)] = ∂v (u∂u − v∂v ) − (u∂u − v∂v )(u∂v )

= u2 ∂u ∂v − u∂v − uv∂v ∂v − u∂v − u2 ∂u ∂v + uv∂v ∂v
= −2u∂v
= ρ(−2x)
= ρ([x, h]),

[ρ(y), ρ(h)] = v∂u (u∂u − v∂v ) − (u∂u − v∂v )(v∂u )

= v∂u + uv∂u ∂u − v 2 ∂u ∂v − uv∂u ∂u + v∂u + v 2 ∂u ∂v
= 2v∂u
= ρ(2y)
= ρ([y, h]).
By the anti-symmetry of the bracket, ρ respects the bracket on the basis elements.
By linearity, ρ respects the bracket in general, and hence it is a Lie algebra homomorphism.
2. By definition, k[u, v]n = hu0 v n , u1 v n−1 , . . . , un v 0 i. There are n + 1 generators which are linearly indepen-
dent, so they are a basis for the subspace and it has dimension n + 1. Consider the action of x, y, h on the
basis elements:
x · un−i v i = u∂v (un−i v i )
= iun−i+1 v i−1

y · un−i v i = v∂u (un−i v i )

= (n − i)un−i−1 v i+1

h · un−i v i = u∂u (un−i v i ) − v∂v (un−i v i )

= (n − 2i)un−i v i

3
 In each case the result is 0 or a homogeneous polynomial of degree n.
Thus, L · k[u, v]n ⊂ k[u, v]n , and so it is a subrepresentation.
1 n 1 n
3. It’s clear that φ(x · v0 ) = x · ( n! u ) = 0 and φ(y · vn ) = y · ( 0! v ) = 0.
Furthermore, we have:
1
hφ(vi ) = (u∂u − v∂v )(un−i v i )
(n − i)!
n − i − i n−i i
= u v
(n − i)!
= (n − 2i)φ(vi )
= φ(h · vi ).

1
xφ(vi ) = u∂v (un−i v i )
(n − i)!
i
= un−i+1 v i−1
(n − i)!
i(n − (i − 1)) n−i+1 i−1
= u v
(n − (i − 1))!
= i(n − i + 1))φ(vi−1 )
= φ(x · vi ).

1
yφ(vi ) = v∂u (un−i v i )
(n − i)!
n − i n−i−1 i+1
= u v
(n − i)!
1
= un−i−1 v i+1
(n − i − 1)!
= φ(vi+1 )
= φ(y · vi ).

We also know that:

k[u, v]n = Span{un−i vi | 0 ≤ i ≤ n, i ∈ N} = Span{φ((n − i)! · vi ) | 0 ≤ i ≤ n, i ∈ N} ⊆ φ(V (n)).

=⇒ φ(V (n)) = k[u, v]n .

So φ is surjective. Both spaces have dimension n + 1, and we know that a surjective linear map in such a
case is bijective. This implies that φ is an isomorphism of sl2 -representations.

Problem 4
Let V be a finite-dimensional vector space. Show that if x, y ∈ gl(V ) are semisimple and commute, then x + y
is semisimple. Give an example of semisimple x, y ∈ gl(V ) such that x + y is not semisimple.

Solution:
Suppose x, y ∈ gl(V ) are semisimple and commute.
Let λ1 , λ2 , . . . , λk be the distinct eigenvalues of x and let Vx,λi = {v ∈ V | xv = λi v}.
By semisimplicity, V = Vx,λ1 ⊕ Vx,λ2 ⊕ . . . ⊕ Vx,λk .
Let v ∈ Vx,λi for some i.
Then x(yv) = y(xv) = λi (yv), and so yv ∈ Vx,λi .
Hence, each Vx,λi is y-invariant. So by Lemma 4.9 (2), the restriction of y to these subspaces is semisimple.
Thus, there exists a basis of Vx,λ1 such that y is diagonal.
Form a basis of V by taking the union of these bases. They are elements of x-eigenspaces, and thus x is also

4
diagonal w.r.t this basis. Let B ∈ GL(V ) be the transition matrix between this basis and the standard basis.
Then B −1 xB and B −1 yB are diagonal. But the sum of diagonal matrices is diagonal and we have:

B −1 xB + B −1 yB = B −1 (xB + yB)
= B −1 (x + y)B

Thus, x + y is diagonal in this basis. So, x + y is semisimple.

For the second part, we are looking for semisimple matrices s.t. x + y is not semisimple. We know from
the first part that x, y must not commute. Let V = C2 . Then consider:
     
1 1 −1 0 0 1
x= ,y = =⇒ x + y =
1 0 −1 0 0 0

A semisimple ⇐⇒ minimal polynomials of A does not have multiple roots, by Lemma 4.9 (1).
But from linear algebra, A ∈ GL2 (C) =⇒ fA (λ) = λ2 − (trA)λ + (detA) is the minimum polynomial of A.
So, x is semisimple ⇐⇒ (trx)2 − 4 detx 6= 0
But trx=1 and detx = −1 and so x is semisimple.
Similarly, (try) = −1 and dety = 0 and so y is semisimple.
However, x + y is in Jordan Normal Form and is not diagonal.
Thus x + y is not diagonalisable and thus not semisimple.

Note: We also have that tr(x + y) = det(x + y) = 0 and so the minimum polynomial has multiple roots.